Hi Ali,
While the numLoadsReady function is indeed not used at the moment, why
would you want to remove it? This code can be useful in the future.
Moreover, the bug is quite easy to correct (you just have to increment
load_idx). So should I provide a patch that correct the bug or one that
remove the code ?
Thanks,
Nathanaël
Le 10/08/2012 18:45, Ali Saidi a écrit :
Hi Nathanaël,
Thanks again for your continued feedback and bug
reporting. We really appreciate it. It looks like you've found another
issue. Both the LSQ unit code and the LSQ numLoadsReady seems to suffer
from this problem. Could you post a patch to review board with all this
code removed that we could commit?
Thanks again,
Ali
On 09.08.2012
09:36, Nathanaël Prémillieu wrote:
Hi all,
I am currently trying
to learn how the O3 cpu model is working. While
reading the code, I
have found this function (in
src/cpu/o3/lsq_unit_impl.hh):
template
int
LSQUnit::numLoadsReady()
{
int load_idx =
loadHead;
int retval = 0;
while (load_idx != loadTail) {
assert(loadQueue[load_idx]);
if
(loadQueue[load_idx]->readyToIssue()) {
++retval;
}
}
return
retval;
}
As the load_idx is never changed, this is an infinite
loop.
Fortunately, this function seems not to be used in the global
code. But
I think leaving a bug even in dead code is not a good
idea.
Thanks,
Nathanaël Prémillieu
_______________________________________________
gem5-dev mailing
list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev
_______________________________________________
gem5-dev mailing list
[email protected]
http://m5sim.org/mailman/listinfo/gem5-dev
